pgml chat blog #914
Merged
pgml chat blog #914
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Really cool post! Kind of an eye opener for me for this use case. |
montanalow
reviewed
Aug 14, 2023
|
|
||
| With its knowledge base in place, now the chatbot links to models that allow natural conversations: | ||
|
|
||
| - Based on users' questions, querying the indexed chunks to rapidly pull the most relevant passages. |
There was a problem hiding this comment.
Should this be a numeric list?
|
|
||
| !!! | ||
|
|
||
| 3. Copy the template file to `.env` |
There was a problem hiding this comment.
Could use cp .env.template .env
|
|
||
| !!! code_block | ||
|
|
||
| ```bash |
There was a problem hiding this comment.
@chillenberger I'm struggling to remember why we need three levels of nesting in markdown, to represent every single code block. Seems like the default style for triple backticks should "handle it".
montanalow
approved these changes
Aug 16, 2023
|
|
||
|
|
||
| # Introduction | ||
| Language models like GPT-3 seem really intelligent at first, but they have a huge blindspot - no external knowledge or memory. Ask them about current events or niche topics and they just can't keep up. To be truly useful in real applications, these large language models (LLMs) need knowledge added to them somehow. The trick is getting them that knowledge fast enough to have natural conversations. Open source tools like LangChain try to help by giving language models more context and knowledge. But they end up glueing together different services into a complex patchwork. This leads to a lot of infrastructure overhead, maintenance needs, and slow response times that hurt chatbot performance. We need a better solution tailored specifically for chatbots to inject knowledge in a way that's fast, relevant and integrated. |
There was a problem hiding this comment.
Suggested change
| Language models like GPT-3 seem really intelligent at first, but they have a huge blindspot - no external knowledge or memory. Ask them about current events or niche topics and they just can't keep up. To be truly useful in real applications, these large language models (LLMs) need knowledge added to them somehow. The trick is getting them that knowledge fast enough to have natural conversations. Open source tools like LangChain try to help by giving language models more context and knowledge. But they end up glueing together different services into a complex patchwork. This leads to a lot of infrastructure overhead, maintenance needs, and slow response times that hurt chatbot performance. We need a better solution tailored specifically for chatbots to inject knowledge in a way that's fast, relevant and integrated. | |
| Language models like GPT-4 seem really intelligent at first, but they have a huge blindspot - no external knowledge or memory. Ask them about current events or niche topics and they just can't keep up. To be truly useful in real applications, these large language models (LLMs) need knowledge added to them somehow. The trick is getting them that knowledge fast enough to have natural conversations. Open source tools like LangChain and LlamaIndex try to help by giving language models more context and knowledge. But they end up glueing together different services into a complex patchwork. This leads to a lot of infrastructure overhead, maintenance needs, and slow response times that hurt chatbot performance. We need a better solution tailored specifically for chatbots to inject knowledge in a way that's fast, relevant and integrated. |
| # Introduction | ||
| Language models like GPT-3 seem really intelligent at first, but they have a huge blindspot - no external knowledge or memory. Ask them about current events or niche topics and they just can't keep up. To be truly useful in real applications, these large language models (LLMs) need knowledge added to them somehow. The trick is getting them that knowledge fast enough to have natural conversations. Open source tools like LangChain try to help by giving language models more context and knowledge. But they end up glueing together different services into a complex patchwork. This leads to a lot of infrastructure overhead, maintenance needs, and slow response times that hurt chatbot performance. We need a better solution tailored specifically for chatbots to inject knowledge in a way that's fast, relevant and integrated. | ||
|
|
||
| In the first part of this blog series, we will talk about deploying a chatbot using `pgml-chat` command line tool. In the second part, we will show how `pgml-chat` works under the hood and focus on achieving low-latencies. |
There was a problem hiding this comment.
Suggested change
| In the first part of this blog series, we will talk about deploying a chatbot using `pgml-chat` command line tool. In the second part, we will show how `pgml-chat` works under the hood and focus on achieving low-latencies. | |
| In the first part of this blog series, we will talk about deploying a chatbot using the `pgml-chat` command line tool. In the second part, we will show how `pgml-chat` works under the hood and focus on achieving low-latencies. |
| 2. Passing those passages to a model like GPT-3 to generate conversational responses. | ||
| 3. Orchestrating the query, retrieval and generation flow to enable real-time chat. | ||
|
|
||
| ## 3. Evaluating and Fine-tuning chatbot |
There was a problem hiding this comment.
Suggested change
| ## 3. Evaluating and Fine-tuning chatbot | |
| ## 3. Evaluating and Fine-tuning the chatbot |
|
|
||
| ## 3. Evaluating and Fine-tuning chatbot | ||
|
|
||
| Chatbot needs to be evaluated and fine-tuned before it can be deployed to the real world. This involves: |
There was a problem hiding this comment.
Suggested change
| Chatbot needs to be evaluated and fine-tuned before it can be deployed to the real world. This involves: | |
| The chatbot needs to be evaluated and fine-tuned before it can be deployed to the real world. This involves: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.